itnewsbot, to tech
@itnewsbot@schleuss.online avatar

AI with a moral compass? Anthropic outlines “Constitutional AI” in its Claude chatbot - Enlarge / Anthropic's Constitutional AI logo on a glowing orange backgr... - https://arstechnica.com/?p=1937794 ⁢

bornach, to random
@bornach@masto.ai avatar

[IBM Technology] on why "hallucinate"

https://youtu.be/cfqtFvWOfg0

itnewsbot, to random
@itnewsbot@schleuss.online avatar

“We must regulate AI,” FTC Chair Khan says - Enlarge / FTC Chair Lina M. Khan testifies during a Senate Commerce, Sc... - https://arstechnica.com/?p=1936486 ⁢

itnewsbot, to gaming
@itnewsbot@schleuss.online avatar

AI vs. Hollywood: Writers battle “plagiarism machines” in union talks - Enlarge / An AI-generated image of "an office copy machine in front of ... - https://arstechnica.com/?p=1936380 &culture ⁢

petersuber, to random
@petersuber@fediscience.org avatar

1/ Conjecture about AI/LLM tools

Find a scientific question with these three properties:

  1. It has a yes/no answer.
  2. It has a history of disagreement. Historically there have been many researchers, experiments, studies, datasets, and arguments on each side.
  3. Recently the question was decisively resolved. The latest research shows consensus, while the past research shows lack of consensus.

,

🧵

bornach,
@bornach@masto.ai avatar

@News4wombats @petersuber
Especially given that tend always include Wikipedia in the training data
https://en.m.wikipedia.org/wiki/Quantum_entanglement

And may even train for many more epochs than with other data sources in order to emphasise its higher credibility.

See
https://youtu.be/E5OnoYF2oAk

Motherboard, to random
@Motherboard@federated.press avatar

Volunteers who maintain the digital encyclopedia are divided on how to deal with the rise of AI-generated content and misinformation.
https://www.vice.com/en/article/v7bdba/ai-is-tearing-wikipedia-apart
-E

itnewsbot, to random
@itnewsbot@schleuss.online avatar

Warning of AI’s danger, pioneer Geoffrey Hinton quits Google to speak freely - Enlarge / Geoffrey Hinton, chief scientific adviser at the Vector Insti... - https://arstechnica.com/?p=1935563 ⁢ -4

delong, to random
@delong@mastodon.social avatar

I Had Been Unaware of How Many People Are Claiming Chat-GPt5 Will Be "Emergent" & Will "Generalize" & Be Nearly "AGI"

I find that I really cannot believe them. Indeed, I am wondering if how many of them are the same people as those who were the crypto hype-grift artists last year…
<https://braddelong.substack.com/p/i-had-been-unaware-of-how-common>
2023-05-01 Mo

bornach,
@bornach@masto.ai avatar

@JoeConst @delong
Marc Evanstein looked at that "Sparks of AGI" video and also read "How Not to Test GPT-3 -
Why doing psychology on large language models is harder than you might think" by Gary Marcus and Ernest Davis,
https://garymarcus.substack.com/p/how-not-to-test-gpt-3

then put through some tests of his own and came to different conclusions regarding the so-called intelligence of

https://youtu.be/d_7EsKcn8nw

bornach, to internet
@bornach@masto.ai avatar

@adamconover speaks to @emilymbender and
@timnitGebru about
https://youtu.be/jAHRbFetqII

itnewsbot, to random
@itnewsbot@schleuss.online avatar

Nuke-launching AI would be illegal under proposed US law - Enlarge / An AI-generated image of a nuclear mushroom cloud. (credit: M... - https://arstechnica.com/?p=1935261 #u.s.government #u.s.senate ⁢ -4 #f#

bornach, to music
@bornach@masto.ai avatar

is still better than me at composing music, but at least I might be able to predict what gravity would do to a tic-tac-toe board

Marc Evanstein puts GPT-4 to the test
https://youtu.be/d_7EsKcn8nw

itnewsbot, to random
@itnewsbot@schleuss.online avatar

ChatGPT now allows disabling chat history, declining training, and exporting data - Enlarge (credit: OpenAI / Stable Diffusion)

On Tuesday, OpenAI... - https://arstechnica.com/?p=1934271 ⁢

ct_bergstrom, to random
@ct_bergstrom@fediscience.org avatar

In our book Calling Bullshit, Jevin West and I provide the following definition.

involves language, statistical figures, data graphics, and other forms of presentation intended to persuade by impressing and overwhelming a reader or listener, with a blatant disregard for truth and logical coherence.

Notice that this is what are designed to do. They are designed with a blatant disregard for logical coherence in that they lack any underlying logic model whatsoever.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • anitta
  • khanakhh
  • mdbf
  • InstantRegret
  • Durango
  • Youngstown
  • rosin
  • slotface
  • thenastyranch
  • osvaldo12
  • ngwrru68w68
  • kavyap
  • cisconetworking
  • DreamBathrooms
  • megavids
  • magazineikmin
  • cubers
  • vwfavf
  • modclub
  • everett
  • ethstaker
  • normalnudes
  • tacticalgear
  • tester
  • provamag3
  • GTA5RPClips
  • Leos
  • JUstTest
  • All magazines
    •